This essay considers the differences between a conceptual account of metaphor (such as that of Lakoff and Turner), in which metaphor is ubiquitous in ordinary thought, and a more limited lexicalist/constituent transfer account (such as that of Tversky and Ortony), in which metaphor is more localized. In the latter view, most of the putative metaphors isolated by Lakoff, et al., are lexicalized and thus literal in ordinary usage. Hogan begins by discussing how different sets of data appear to favor one or the other account. He goes on to argue that a modified version of the lexicalist/constituent transfer account can account for all sets of data, including those that initially appear to favor conceptual metaphor. Moreover, the lexicalist/constituent transfer account can do this through a parsimonious cognitive architecture involving algorithmically well-specified use of standard cognitive mechanisms (prominently priming). [P.C.H.]